Você está na página 1de 48

NETHERLANDS

GEODETIC

PUBLICATIONS O N GEODESY
VOLUME 3

COMMISSION
NEW SERIES
NUMBER 4

THE THEORY OF DISPERSION


APPLIED TO ELECTRO-OPTICAL
DISTANCE MEASUREMENT AND
ANGLE MEASUREMENT
by

J. C. DE MUNCK

1970
RIJKSCOMMISSIE VOOR GEODESIE, KANAALWEG 4, DELFT, NETHERLANDS

PRINTED BY W. D. MEINEMA N.V., DELFT, NETHERLANDS

CONTENTS
page

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

List of units and symbols


Summary

Chapter 1 Introduction

. . . . . . . . . . . . .

Chapter 2 The group propagation time of an amplitude-modulated signal

. . . .

Chapter 3 The influence of an inhomogeneous medium on electromagnetic distance


measurement using one carrier wave . . . . . . . . . . . . . . . . 12

17

Chapter 5 Electromagnetic distance measurement on two optical wavelengths and


one micro wavelength . . . . . . . . . . . . . . . . . . . . . .

19

Chapter 4 Electromagnetic distance measurement on two optical wavelengths .

Chapter 6 A survey of the non-instrumental inaccuracies of electromagnetic distance


measurements . . . . . . . . . . . . . . . . . . . . . . . . . 22
6.1 General

22

6.2

. . . . . . . . . . . . . . . . . . . . . . . . . .
Explanations to table 8 . . . . . . . . . . . . . . . . . . .
General remarks and conclusions on table 8 . . . . . . . . . .

22

. . . . . . . . . . . . . .

26

. . . . . . . . . . . . . . . . . . . . . .

29

. . . . . . . . . . . . . . . . . . . . . . . . . . .

30

. . . . . . . . . . . . . . . . . . . . . . . . . . . .

31

6.3

Chapter 7 The dispersion for angle measurements


Chapter 8 Concluding remarks .
References
Tables

. . . . . . . . . . . . . . . . . .
The Lorentz-Lorenz equation for the refraction index . . . . . . . .

Appendix I Some values used in this paper

I1

24

39
43

111 The error introduced if an electromagnetic distance measurement is calculated with the group refraction index . . . . . . . . . . . . . . 44
IV The influence of errors in estimating the fictitious temperature te . . . 45

V The influence of errors in the group refraction index (ML.


S&) and in
the dispersion (dALfi) . . . . . . . . . . . . . . . . . . . . . . 47

LIST O F U N I T S A N D SYMBOLS

unit of length:
metre
unit of temperature: centigrade
torr
unit of pressure:

Symbol
A

c,
C
C

D
e
Fa,

F*

9
9

H
I,, I,, I,,, I,,, I,,
I,,, I,,
KL, KM

unless otherwise stated

meaning
starting point of a light- or radio path
end point of a light- or radio path
humidity correction on refraction angle
light velocity in vacuum
index for carrier wave
= G,/A,G, dispersion factor
= (El - I)/ALfi
humidity function of a particular place
functions of a particular place
phase-dispersion function for dry air
group-dispersion function for dry air
as G and G,but for the Lorentz-Lorenz equation
= 111, the inverse of the wavelength in vacuum
index for using the group refraction index
geometric elevation of B in A
integrals of meteorological conditions on the X-axis
integrals of meteorological conditions on the X-axis
factors to calculate a distance from measurements on
three wavelengths
= R anL/aZ, refraction coefficient for visible light
= R anM/aZ, refraction coefficient for microwaves
index for light
index for microwaves
denominater of KL and KM
(phase) refraction index
group refraction index
total pressure of the air in torr
partial pressure of water vapour in torr
partial pressure of CO, in torr
(6.38. 106 m), ray of curvature of the earth in m
geometrical (straight-line) distance between A and B
in chapter 2: time
other chapters: temperature
particular average temperatures introduced in 6, and 8,

Symbol
UA, U,
U0
X, y, z

meaning
Page
vibration function in A or in B
8
mean amplitude of U,
8
carthesian coordinates; origin in A , X-axis through B, Y-axis
parallel to horizontal plane in B
14,26
extinction between A and B
8
phase-dispersion function for humidity correction
12,31
group-dispersion function for humidity correction
14,31
operator for a quantity on the wavelength I , minus the quantity
for I , (both optical wavelengths)
17
operator for a quantity on a micro wavelength minus the
quantity on the optical wavelength I ,
17
operator indicating a small variation
22,36,37,47
correction if using the Lorentz-Lorenz form
43
13
small quantity, independent of the position
observed elevation of B in A
26
functions allowing for the difference between radio waves
and light waves
19,42
effect of dispersive extinction on travelling time
9
10
wavelength in vacuum
a function of the position (p z l or p
13
1)
= 3.14.. .
10
function of the local temperature and total pressure
12,31,39
(Q is roughly proportional to the density of the air)
= c X 2, optical path length
10
= c X 2,, measured path length for a group
10
path length calculated with n"
44
retardation or travelling time
8, 10
second order terms
14
angular frequency ( = 2x X frequency)
8
angular frequency of modulation
8
angular frequency of carrier
8
index belonging to primary optical wavelength
17
index belonging to secondary optical wavelength
17

SUMMARY

A general expression is derived for the group propagation time in a dispersive inhomogeneous medium. This expression is more exact than the usual concept of the group refraction
index.
The above mentioned expression is applied to an extension of the theory of M o ~ r z .
MORITZgives the propagation time and the refractive angle in an inhomogeneous medium
as a power series expansion. So first order- and second order corrections or errors are
derived for electromagnetic distance measurement on one, two and three wavelengths and
for angle measurement on two wavelengths.
For the E.D.M. the following influences are considered: the first order dry air correction
for the refraction index, the first order humidity correction, and the dry air-, humidity-,
and mixed curvature corrections. For the three-wavelengths method the different temperature dependence for light- and for microwaves and the inaccuracies in the dispersion formulae for the air are also considered.
For angle measurements, where the dispersion effect is extremely small, only the first
order corrections for dry air and for humidity are considered.
For E.D.M. an accuracy of a few parts in 109 seems to be obtainable provided that the
refractive index of air as a function of the wavelength, the temperature, the pressure and
the humidity are known with sufficient accuracy, and provided that the instrumental errors
are sufficiently small.
For angle measurement on two optical wavelengths it will be very difficult to obtain a
sufficient instrumental accuracy because of the small phase dispersion and because of the
serious influence of very local effects in angle measurements.

Chapter 1
INTRODUCTION

The accuracy of electromagnetic distance measurement (E.D.M.) is seriously influenced by


local variations of the refractive index of the air. To correct for these variations one should
know the average group refraction index for E.D.M. or, for measuring angles, the average
value of the variations of the phase refraction index perpendicular to the path. In a dispersive
medium, i.e. a medium in which this refraction index depends on the wavelength, the
average values can in first order approximation be eliminated by measuring the distance
(or the angle) simultaneously on two wavelengths, provided the measurements can be
executed with sufficient accuracy.
To correct also for the average composition of the air (humidity) it may be useful to
measure on three wavelengths instead of on two.
The possibilities and the limitations of these methods are considered by extending the above
mentioned first order theory.
With respect to E.D.M., a general expression. is derived for the group travelling time
(equation 5a). In order to derive detailed expressions for calculating the geometric distance
from travelling time measurements and from meteorological measurements, the general
expression is applied to a theory given by MORITZin [l]. In this theory MORITZexpresses
the geometric distance in a measured travelling time and in integrals of the refractive index
and its derivates along a straight line. This theory may be used for E.D.M. in an inhomogeneous medium where the variations of the refraction index are small and gradual.
The theory derived in this paper is aimed at the estimation of errors caused by different
physical sources in E.D.M. using one, two or three wavelengths. Numerical results are
given in table 8.
Similar considerations lead to an estimation of the errors in agle measurements (chapter 7). Here only a first order approximation is used because the effects are small. Moreover
the measurement of the refraction angle is extremely difficult.
For many years different investigators have tried to use the dispersion method to measure
or to eliminate the refraction angle: [2], [3, pag. 811, [4], [20]. Up to now the success does
not seem to be very great.
In the case of E.D.M. dispersion measurements are better applicable because the group
dispersion is considerably larger than the phase dispersion (about three times as large).
Moreover by using photo-electric methods, as is done in E.D.M., one can spread the wavelengths more than in the visual case. Promising experiments have recently been described by
J. C. OWENS[5] and by M. T. PRILIPIN[6].

Chapter 2
T H E G R O U P PROPAGATION T I M E O F
AN AMPLITUDE-MODULATED SIGNAL

In electronic distance measurement the travelling time is measured, or more exactly the
phase difference of a modulation of the carrier wave between starting and end point. If the
medium is dispersive (i.e. the propagation velocity changes with the frequency) the travelling
time of the modulation (group travelling time) differs from the travelling time of a strictly
monochromatic wave (phase travelling time).
In this chapter an expression will be derived giving the group travelling time for an
amplitude-modulated wave received by a quadratic detector (e.g. photo-multiplier).
Suppose in A (the starting point) there is an amplitude-modulated vibration with an angular
frequency of the carrier o, and an amplitude U, .(l +m sin w, t), in which m is the degree
of modulation and o, (<< W,) the angular frequency of the modulation. This vibration may
be written as [7, section 7 par. l ] :

or as the sum of three harmonic vibrations:

Suppose in the end point B we have a vibration U,(t) related to UA(t)by linear differential
equations. This assumption will do for all cases of electronic distance measurement. Then
U,(t) may be written as the sum of three components with the same frequencies as in UA(t),
however with retardations z(o) and losses B(o) which may be considered as linear functions
within the narrow frequency band o, -W, < o < o, + W,. So one can state:

dB
B(%-W,) = B-osdw

dB
P(oc+oJ = B+ws- d o

with :
z = ~(0,)

9 = ~(w,)

dT
do

2 (F)
=

for w = w,

. . . . . . . . . .

(2)

THE THEORY OF DISPERSION, ETC.

So the vibration in the end point becomes:


UB(t)= p(w,) . U, sin {w,t - o; z(w,)}

UB(t)= p U, sin (w,t - o , z )

For phase measurement of the modulation of a light signal, use is made of a squaring
detector, for example a photo-multiplier. In such a detector the signal is filtered so, that
only the low frequency components can be observed ( w of the order of magnitude of W , ) *).
After some calculations one can write for the relevant terms of U i ( t ) :

~ i ( t=) ... +/? m U . COS w, -- sin w,t - o,z - w,o, -- d


dro

or with

The term @ of the phase of the first component appears to be extremely small for electrooptical distance measurement. This is demonstrated in the following unfavourable case:
*) This is not the case with interference of carrier waves in the Vaisala base measurement method.

PUBLICATIONS ON

GEODESY, NEW SERIES, VOL. 3,

NO.

o, = 2x. 10i5 radianslsec (carrier wavelength z 0.3 pm)


o, = 2x0109 radianslsec (modulation wavelength z 30 cm)
z = 10-3 sec (distance % 300 km)

for atmospheric air if n is the (phase) refraction index at a circular


frequency o, and A, the (vacuum) wavelength.
With these "unfavourable" values one finds:

So if the dispersion of the attenuation*)


tg is extremely small indeed.
Comparing the phases of the right hand terms of (4) with the phase of the modulation of
(1) and neglecting @ one finds the retardation z, of the modulation, i.e. the measured
travelling time: see equation (5a), where the index c for the carrier wave has been omitted.
Equation (5) can also be written as (5b), (5c) and (5d), where use is made of:
the optical path length a or a,, defined by .r or z, multiplied with the light velocity in
vacuum c;
- the wave number g (6c);
- the wavelength in vacuum A (6d).
-

z and a are values for the (monochromatic) carrier wave, the index m is used for the measured
values on the modulation.

Definitions :
(6~) g

o
(= 1/A)
271c

--

(6d)

1 = 2zclw

. . . . . . . (6)

*) This dispersion may be caused by absorption in the atmosphere or in any optical or electronical filter,

or in the directional sensitivity of the instrument.

THE THEORY OF DISPERSION, ETC.

11

The relation (5d) is very similar to the expression used in the resolution of the U.G.G.I. [8],
giving the "group refraction index" E as a function of 1 and the phase refraction index n:

Both relations are physically identical if it is supposed that the electromagnetic (or other)
waves follow a straight-line path between A and B. This is the case normally found in textbooks about group propagation.
Note: For radio waves in the lower atmosphere the dispersion is so small that phase propagation time may
always be used in this case.

Chapter 3
THE INFLUENCE O F AN INHOMOGENEOUS MEDIUM
O N ELECTROMAGNETIC DISTANCE MEASUREMENT USING
O N E CARRI'ER WAVE

The general principle of the geometric optics is a relation between the phase refraction
index n in a point and the optical path a between some origin and that point.

This relation, corresponding to the well-known principle of Fermat, may be derived from
the field equations of Maxwell for a purely monochromatic wave in an isotropic inhomogeneous medium which does not change considerably over a distance of one wavelength
[9, ch. 1111.
The refraction index n in the lower atmosphere may be written as a function of the place
and of the wavelength, if time-effects are not considered. OWENS[l01 gives probably the
most accurate expressions for n as a function of the wavelength, the meteorological conditions and the composition of the air for visual light in the atmosphere. In our paper however
the much simpler formulae of EDLBN[11, eq. 22, 12 and l ] will be used for visual light.
For radio waves the equation of ESSENand FROOME
will be used [12, eq. 21. Only when
considering the errors introduced by these equations (chapter 6 and appendix 11), the
OWENS'expressions are introduced.
When using the equations of EDLBNand of ESSENand FROOME,
n may be written as:

where:

n is considered as a function of the position and of g,


g is the inverse of the wavelength in vacuum,
G and f are only functions of g (for light) or constants (for radio waves),

is a function of the local temperature and the total pressure of the air (Q is
roughly proportional to the density of the air), and
e is a function of the local partial pressure of the water vapour and of the local
temperature and total pressure (for dry air e = 0).
@

Hence for light waves, and also for radio waves, Q and e are only functions of the place.
G, f , @ and e will always be chosen so, that and e never become much larger than unity.

THE THEORY OF DISPERSION, ETC.

13

Some values are calculated in appendix I. G appears to be about 300. 10-6 for light- and
for radio waves. r, for radio waves (microwaves) has about the same magnitude, but for
light T , z - 4 . 1 0 - ~ . The more complete values are mentioned in table 1.
In [l] MORITZgives an approximation for calculating the optical path and the refraction
angle from (8), if n is known as a function of a particular place. MORITZstates:

where E ( z 0.0006) is independent of the place, and p ( z 1 or c 1) is a function of the


place only. (The dependence on the wavelength is not mentioned in [l]).
MORITZwrites the optical path a from an origin A to the point B as a power series
in E:

...

a = S+EF,+E~F,+

where S is the geometric distance from A to B, and F,, F,, ... are functions of the place of B.
This series is only useful if n does not change too irregularly with the place. In cases of
mirage, duct or reflections - when more than one light- (or radio) ray exists between A and
B - and in the case of strong turbulences, the series does not give always a useful model.
Using this series MORITZderives an expression for the geometric distance between A
and B. This expression contains the refraction index and its partial derivates in all points
of a straight line between starting point and end point. If this line is running through extra
disturbed regions (near the ground) or a fortiori through the ground one must take for n
and its derivates an extrapolation of the n field nearer to the physical rays.
For the refraction angles the elevation and the bearing of the starting point A, seen from
the end point B, are written as power series in E. Here the same restrictions may be mentioned.
Tf in the whole relevant space n is multiplied by any constant factor, the refraction angles
do not change, and the optical paths are multiplied with the same factor. So the theory is
not only true if n z 1, but in any case provided the relative variations of n are small.
In this paper the theory of MORITZis accepted.
Our equation (9) is consistent with the statement of MORITZ(10) if:

Using the results of MORITZ[8, eq. (3), (4), (8), (1 5), (13')], the optical path length of a
monochromatic wave is obtained after some calculations (12a). Introduction of (1 1) gives
(12b).

a = ~ + G z , + ~ I ~ - + G ~ I ~ ~ - G ~. z. , .~ .- .+ .~ .~ .I ~. ~.

where :

. . . (12b)

14

PUBLICATIONS ON GEODESY,NEW SERIES, VOL.

3, NO. 4

and where :

S is the geometric straight-line distance between the starting point A and the end point B,
X, Y, Z are carthesian coordinates with the origin in A and the X-axis through B, and the
Y-axis parallel to the horizontal plane in B *),
- all integrals are taken along the X-axis, i.e. e, e, and its derivates are taken for Y = Z = 0,
- all terms of third- and higher degree in G and f are neglected.
-

The equations (12a) and (12b) give only a monochromatic solution of the wave propagation. In electromagnetic distance measurement however the optical pathlength a, of a
modulation is measured. This value may be found by introducing (Sc). One finds:

S = ~,-GI,-~"I,+Q

. . ... . . . . . . . . . . . . . . . . .

(14)

with :

and:

In the expression (14):

is the wanted geometric distance between A and B,


follows directly from the measured travelling time by multiplying with c,
a,
( - GI,) is the correction for the (group) velocity in dry air,
(-PIe) is the correction on the term (-GI,) for the humidity, and
52
contains second order corrections.
*) With this choice of the axes the refraction in the Y-direction may be called lateral refraction, which nor-

mally will be much smaller than the (vertical) refraction in the Z-direction.

15

THE THEORY OF DISPERSION, ETC.

In order to find some quantitive statements on the different influences on the resulting
distance S, a simplified model will be used:
simplified model

a The distance is "horizontal", i.e. starting- and end point are on


one level surface.
b The values of Q and e are constant on this (spherical) surface.
c The lateral refraction is negligible, i.e. ae/aY = 0 and delay = 0.
d ae/aZ and ae/aZ are constant in the relevant space.
In this model the equation (18) for the integrals of (13) is easily found *).

For the relative correction on the direct measured optical path length one finds:

*) Derivation of the equation (18)

The intersection of the (spherical) level surface through the end points A and B with the plane Y
is described by the circle:

=0

x+Z~-XS+Z~~R~
=-0,
S~
if R W 6.38.10Bm is the ray of curvature of the level surface. Since R is not much smaller than the
usual rays of curvature of the light- or radio path, S2may be neglected in the approximations of (12a)
and (12b). So one finds for the level curve (where e and e have the constant values p, and e,):
If ae/aZ and ae/aZ are constant in the relevant field, one finds for the values of
(that is along the straight line A-B):

a@x ( s - X )

e = e ~ - - -az

2~

and

e,

and e along the X-axis

ae X(S-X)
2~

--

az

Substituting these values and the constant values ae/aZ and ae/aZ in (13) and omitting the index 0 the
equations (18) are obtained.

16

PUBLICATIONS ON

GEODESY,
NEW SERIES,
VOL. 3,

NO.

or with very good approximation (see appendix 111):

where:

1- E +

s2 afi
s2

----

12R

az

(g)l

. . . . . . . . . . - (19b)

R is the radius of the level surface through A and B; and


E is the group refraction index defined by (7).

Based on the literature and on own estimates some values of the air refractivity and of
meteorological conditions are derived in appendix I and compiled in table 1. The meteorological values are indicated for "normal conditions" (usual in moderate climats) and for
exceptional conditions (seldom occurring in different climates, but not strictly extremes).
With these values of table 1 the different terms of (19a) have been calculated, see table 3.
The physical meaning of the terms is indicated in table 2.
It may easily be seen that the second order terms can become important, particularly for
the longer distances (tens of kilometres). It is also clear that the humidity may not always
be neglected in electro-optical distance measurement.

Chapter 4
ELECTROMAGNETIC DISTANCE MEASUREMENT
O N TWO OPTICAL WAVELENGTHS

By this method two independent measurements are executed:


1 The optical path a,, measured on the optical wavelength A,, and
2 The difference ALa, = a,, -a,, between the optical paths on two wavelengths A, and A,.
This difference can often be measured with a much higher precision than the path a,,
itself.

Owing to the dispersion of the air it is possible to calculate the distance S if all values
in (14) are known or negligible except S and I,. Writing down (14) for A, and for A, one
finds after some calculations :

where:

D = G,/A,G can be called "dispersion factor";


the operator A, indicates the value of the next quantity for A, minus the value
for A, ;
index L indicates an optical wavelength for which the equations (a) and (c) of
appendix I hold;
the index M will be used in the next chapter to indicate microwaves (appendix I
equations (b) and (d));
the indices 1 and 2 indicate the (optical) wavelengths A, and A,.

,,

a,, and A,a, are directly measured. The values of G,, f etc. can be calculated with high
precision from laboratory measurements [10:) [l l]. The dispersion factor D appears to be
about 10 for two wide-spread visible wavelengths A, and A,.

The different meteorological effects on the measurements will be estimated using the
model (17). Substituting (15) and (18) into (20) one finds the relative difference between
the distance S and the distance (a,, - DA,a,) calculated from the measured path lengths
a,, and ALa, :

18

PUBLICATIONS O N GEODESY, N E W SERIES, VOL.

3,

NO.

or with good approximation (see appendix 111):

With the values of table l the terms of (21a) have been calculated, see table 4. The physical
meaning of the terms is indicated in table 2. Comparing the case of two optical wavelengths
(table 4) with the measurements on one optical wavelength (table 3) one sees:

1 The first order dry air correction is non-existent for the two wavelengths.
2 The humidity corrections are bigger for the two wavelengths than for one wavelength.
3 The second order dry air correction is smaller for the two wavelengths in most of the cases.

Chapter 5
ELECTROMAGNETIC DISTANCE MEASUREMENT
O N TWO OPTICAL WAVELENGTHS A N D
O N E MICRO WAVELENGTH

For this method three independent measurements are executed :


1 The optical path a,, on the optical wavelength A,.
2 The difference ALa, = a,,-a,,
between the optical paths on two wavelengths A , and
A,, and
between the optical paths on a microwave A, and on
3 The difference A,a, = a,,-a,,
the optical wavelength A , .
Like in chapter 4 the differences may often be measured with a much higher precision than
the optical path a,, itself. Owing to the dispersion of dry air for optical waves and owing
to the difference in velocity between optical waves and radio waves in water vapour, it is
possible to correct the measurements for the mean refraction index of dry air and for the
mean humidity. However some difficulties arise on account of the different temperaturehumidity dependence of the refraction indices for optical- and for radio waves. Also the
well-known reflections and other double path effects on radio waves might give much
trouble, which may however, at least in principle, be eliminated by instrumentation and
measuring methods.
To discuss the calculations for this three-wavelengths method equation (14) is written
for A,, A, and A, as:

s+~,I,~+T",I,~

= a,,

e2~eL
+f 2 ~ e =L

+Q,

+ QZ
S + GMIeM+PMleM= amM+ QM
S+

bmZ

. . . . . . . . . . . . . . . . . (22)

The straight-line integrals Z are not identical for light- and for microwaves because Q and
particularly e are different functions for the two types of waves. For Q the difference is
very small but for e it may be quite significant.
As the ratio's 8, =
and 8, = eM/eLdo not change very much with the meteorological conditions along the path, it is useful to state IeM/IeL= 8, and ZeM/IeL= 8, for some
mean conditions. From appendix I, equation (I.l), follows with good approximation:

where t, is some average temperature along the path. For t, an estimation of the mean
temperature can be introduced. The approximation (23) will be used for accuracy considerations; for the calculations of the distance from measurements the more accurate first forms
of (1.1) may be used (see appendix I).

20

PUBLICATIONS ON GEODESY, NEW

SERIES, VOL. 3, NO. 4

With the values 8, and 8, the equations (22) may be written explicitly in the measured
quantities a,, ALa, and AMa,. Equation (24) is then obtained, from which the wanted
distance S may be solved (25). The integrals I,, and I,, may of course also be solved.

S = a,, - K L A L ~-, KMAMa,+ Q , - KLALSZ- KMAMQ . . .

. . . . . .

(25)

with :

G -DzlO
-W---d L c . ( e e ~ M - r " l ) - ~ L P . ( e , ~ M - cA,L)c

G,FMe,
- r",GMO,
K L = -KM -

~,.AL~--G,.ALF

A,G.(e$,

- F,) - AL7:.(8,G, -

c,)

m
m

Plc2-GlP2

-0.02

~,T,A~G

K, and KM are factors that depend - apart from the small influence oft, and the very small
influence of t, - only on the used optical wavelengths R , and R,. The numerical values of
K, and KM are calculated from table 1 for R , = 0.625 pm and R, = 0.3636 pm.
The different meteorological effects on the distance calculated from the measurements
will be estimated using the model (17). Substituting (15) with (18) into (25) one finds the
relative difference between the distance S and the distance (a,, - KLALa,- K,AMa,) calculated from the measured path lengths a,,, ALum and AMa,:

THE THEORY OF DISPERSION. ETC.

21

or with good approximation (see appendix 111):

Note: Since for microwaves dG/dg


for these waves.

dr/dg

=0

(no dispersion), the wave number g disappears in (27a)

With the values of table l the different terms of (27a) have been calculated, see table 5.
The physical meaning of the terms is indicated in table 2. Comparing the case of two
optical- and one radio wavelength with that of two optical wavelengths only, one sees:

1 Only the higher order terms exist for the 3-l-method.


2 The dry air second order term is essentially equal for both cases.
3 The second order humidity effects may be greater for the 3-l-method.

Chapter 6
A SURVEY O F THE NON-INSTRUMENTAL INACCURACIES
O F ELECTROMAGNETIC DISTANCE MEASUREMENTS

6.1 General
In this chapter the non-instrumental inaccuracies are estimated for the different methods
of E.D.M. considered in the preceding chapter. The results are complied in table 8. These
values have been calculated form the assumed inaccuracies in the meteorological measurements and in the refractivity of the air from the tables 6 and 7 respectively. For the
"favourable" and "typical" cases in table 8 the "typical" values of table 3, 4 and 5 were
taken, for the "unfavourable" cases high absolute values of these tables were chosen.
Most of the values of table 8 may more or less be interpreted as standard deviations; the
influences of the formulae (row 7, 8 and 9) are given as tolerances. The "unfavourable"
cases give rather exceptional, but not extreme values.
With regard to the measurements the following assumptions are made:
1 One measurement takes about 15 to 30 minutes measuring time.
2 Careful measurements of temperature, barometric pressure and humidity are executed
near both ends.
3 Vertical angles are measured at both end points in order to find the vertical density
gradient of dry air a e / a Z . See SAASTEMOINEN
[13].
6.2 Explanations to table 8
RoW 1

The influence of errors in the density function


one wavelength.

Q,

occurring only in the case of

Partial differentiation of (19a), neglecting the higher order terms, gives:

as
ae

- --

sG
l+Ge+Pe

=sG

for

E+1.

So the relative variation of S becomes:

Substitution of the figures from the tables 1 and 6 gives the values in table 8.
R oW 2

The j r s t order influence of the humidity.

Calculated analogously to row 1 but with the values of e and Se from the tables 1 and 6.

THE THEORY OF DISPERSION, ETC.

Row 3

23

The effect of errors in the estimation of the density gradient @ / a Z for dry air.
(The accuracy of the third term of table 3 or the second term of table 4 or thejirst
term of table 5).

The values are calculated assuming 6 ( a ~ / a Z ) / ( a ~ / a=Z0.1


) (table 6), taking the values
for G, P, D, K, etc. from table 1 and with a~/az
from table 6. For a distance of 100 km
however the errors for the "unfavourable" cases are assumed to amount to only twice the
errors for a distance of 30 km.
R oW 4

The second order humidity influence. (The total accuracy of the 4th and 5th term
in table 3 or of the 3rd and 4th term in table 4 or of the 2nd and 3rd term in table 5).

For 6(ae/aZ)/(ae/dZ)unity has been substituted because it is hardly possible to estimate


ae/aZ. So for the values in table 8 for "typical" and "unfavourable" cases the sums of the
relevant terms are taken from the tables 3, 4 or 5. For the "favourable" cases the errors
are supposed to be 25% of the typical values. For a distance of 100 km the errors for the
"unfavourable" cases are assumed to amount to only twice the errors for a distance of 30 km.
If ae/dZ can be calculated from measurements of the humidity gradients, the values of
row 4 may possibly be reduced by a factor 10.
R o w 5 a n d 6 The influence of errors in estimating the temperature t,. This temperature t,
is necessary to account for the dzjierence between the temperature dependence of
the refractive index for light waves and radio waves (essentially for the water
vapour).

t, is assumed to be known with the same accuracy as the mean air temperature (table 6).
The temperature t, may be calculated from temperature measurements at the end points
or from I,,, calculated iterativally from the equations (24)*). The inaccuracies caused
by errors in the accepted value for t, are developed in appendix IV.
R o w 7 , 8 a n d 9 The influence of errors in the formulae giving the refraction index of air
as a function of pressure, temperature, humidity, wavelength, etc.
The error 6 s in the calculated distance is found from the equations (19b), (21b) and (27a)
or (27b). See appendix V. 6 s can explicitly be written as a function of the errors in the
group refraction indices and in the group dispersion 6&, an", and 6ALn".Numerical values
for these errors are taken from literature (see table 7 ) assuming a moderate humidity
( e = 0.1, i.e. p, = 7.2 torr) for "favourable" and "typical" conditions and a very high
humidity ( e = 1 , i.e. p, = 72 torr) for "unfavourable" conditions.
*) Note on the calculation of t , from (24).
The temperature t , is estimated as a first approximation. 0, is calculated with the most accurate form

of (1.1). I,L is calculated from the 2nd and the 3rd equation of (24), neglecting the Q-terms and assuming
t, = t , . Now er, is found with (18): er, = I,r,/S. With appendix I, equation (c), t may be found if the
air pressure p is known. Eventually iteration is possible.
This method seems to be more accurate than the direct measurement of t becausep may be measured
with a good accuracy. However there may be a significant discrepancy between the calculated t and
the wanted t,.

24

PUBLICATIONS ON

GEODESY,
NEW SERIES,
VOL. 3, NO. 4

R o w 10 Errors caused by deviations of CO2-contents of the air.


Variations of the CO2-contents of the air are described in [l41 and roughly in [15]. The
following values are assumed for the mean partial pressure p, and for its standard deviation:

1
1

mean partial
Pressur; of CO2
p,

0.25 torr

standard deviation

1 favourable or typical 1
1

0.02 torr

unfavourable

0.08 torr

The influence on the refraction indices and on the dispersion is determined by

Considering the influence of CO, as errors in the refraction index, the values of row 10 are
found by introducing the figures from the above small tables into the equations (a), (b)
and (c) of appendix V.
R o w 1 1 The uncertainty in the light velocity in vacuum, giving a constant deviation in the
scale.
If the distance is not expressed in "light seconds" but in internationally defined metres the
uncertainty in the light velocity (table 7) may give a corresponding scaling error in the
distance.
6.3 General remarks and conclusions on table 8

1 If the mean value is taken from a number of distances measured in different weather
conditions, the meteorological effects on this mean value will tend to decrease from the
"unfavourable" values to "typical-" and from "typical-" to "favourable" values. This,
however, will hardly be the case for the errors in the formulae (row 7, 8 and 9). For
methods more independent of visibility conditions (microwaves) the gain may be
relatively high.
2 An error in the velocity c of light in vacuum means a constant error in the scale for all
radar-type measurements. Descrepancies will only occur if the measurements are
compared with very accurate distance measurements independent of c (invar wires,
VAISALA
bases).
3 The inaccuracies from the optical formulae (row 7 and 8) are partly constant errors
in the scale for electro-optical distance measurement. So the inaccuracies in the shape
of geodetic configurations will often be a factor 2 better than the values suggested in
the rows 7 and 8.

THE THEORY OF DISPERSION, ETC.

25

As already mentioned in the introduction of MORITZ'theory, ambiguous optical paths,


such as reflections, ducts, etc., may give rise to errors. Although these errors may be
quite important, even for optical wavelengths, table 8 does not account for these effects
because a quantitative description is very difficult and because the effects can be limited
by appropriate design of the instruments and the methods (narrow beams, diversity
in place or in frequency, etc.).
The influence of the CO,-contents and the errors in the formulae of ESSENand FROOME
(row 10 and 9) will nearly always be small compared with other effects. Also the influence of 8, (row 5 and 6) will seldom be important.
For electro-optical distance measurement on one wavelength the effect of Q (row l),
that is essentially the accuracy of the temperature, is normally by far the most important
source of errors. In "unfavourable" cases however on distances of more than 10 km
the influence of the vertical refraction may become even more important (row 3 and
to a less degree, row 4).
For microwave distance measurement (one A) usually the mean humidity (row 2) is
the dominating source of errors. For distances longer than 10 km with great humidity
gradients the errors from row 4 (aelaz) may become significant.
For two optical wavelengths the errors in the formulae (row 7 and 8) are important
and also the humidity of the air (row 2). For distances longer than 3 km also the
vertical refraction may be significant (row 3 and to a less extent, row 4).
For the method with three wavelengths the influence of e (row 2) disappears but the
other factors are not very different from the two-wavelengths method.
With one optical wavelength an accuracy of l ppm should normally be obtainable,
with one micro wavelength about 4 ppm, with two optical wavelengths 0.15 ppm and
with three wavelengths a somewhat better accuracy. The gain of the third wavelength
seems very small. However the influence of the humidity (row 2) may increase considerably if only two wavelengths are used, and the gain will be much higher if the
formulae for the refraction index (row 7 and 8) are better known.
The errors in the measurement of the optical paths (a,, ALa, and A,u,) i.e. the instrumental- and the reading errors etc. are not considered in the above.
The influence of these errors is found by differentiating the calculated distance S
partially in the equations (19b), (21b) and (25). Neglecting the higher order terms
one finds with n" + 1 :
Table 9

method
one wavelength
two wavelengths
three wavelengths

So an error in the measured optical path a, enters directly in the result, an error in
the optical difference ALa is multiplied with (-D) and an error in A,a is decreased in
the result to (-KM) times its value.

Chapter 7
T H E DISPERSION F O R ANGLE MEASUREMENT

Roughly analogue to chapter 4 the theory of MORITZ[l] may be used to calculate the angle
of refraction from measurements of the directions on two wavelengths. For angle measurements however the group travelling time does not enter into the problem: only the phase
refraction index is needed.
Because the effect of the dispersion on the direction is very small only thefirst order effect
on vertical angles will be treated, although there is no real difficulty in deriving the theory
for higher order terms and for bearings.
Suppose a light source B has in the point of observation A a geometric elevation H
(see figure 1).
2,

Figure 1 The elevation of B in A


Rectangular Carthesian coordinate system:

B light source
A observer
H geometric elevation of B in A
7 observed elevation of B in A

Origin in B
X-axis through A
Y-axis perpendicular to the vertical in B

According to MORITZ[l, equation 23, second] is then:

where:

the axes are taken in accordance to figure 1,


an/aZ is approximately a vertical component for a nearly horizontal direction B A ,
the integral is taken along the X-axis.

Introducing (9) one finds:


H

=V

I,,

+ GI,, +I'I,,

with:
1
S

S a@
,S az XdX
--

In this equation only G and

and

I,,

1
S

ae

- S -- XdX
S

,az

r and the measured elevation rj depend on the wavelength.

27

THE THEORY OF DISPERSION, ETC.

So writing down the equation for two wavelengths and taking the difference one finds:

H
0

= ?l
=

+G,I,z+~lIez

AL? + (ALGY,,

with:

ALG

+ (ALT)I,,

G,-G,,

ALT

T,-T,

and

A,?

or eliminating I,,:

G
H = q1 - L A ? + C ,
AG

. . . . . . . . . . . . . . . . . . . . . .

(30)

with :

In equation (30) v, is measured directly, A,? may be measured, although a sufficient precision is difficult to obtain. G , , ALG, T , and ALT can be calculated from formulae for the
refraction index (e.g. appendix I equation (c). I,, however can at best roughly be estimated
from the weather conditions. Since T is very small for (near) optical wavelengths the
humidity correction will in general be neglected.
Table 10 shows that G,/ALG is rather high, particularly for visual observations*) (for
example 1 , = 0.65 pm and 1 , = 0.47 pm). Gl/ALG for phase velocity appears to amount
to thrice the comparable quantity D for group velocity (see last column of table 10). So
the angle difference A,? should be measured with very high accuracy because errors in
A? are multiplied with Gl/ALG.
Table 10. The coefficient of A,? in equation (36)

In order to estimate some numerical values for the refraction angle for the humidity correction, ae/aZ and ae/aZ are supposed to be constant along the X-axis. In this case I,, and
I,, can be written as:

ae

and

ire
1,, = ' E
S

. . . . . . . . . . . . .

(32)

From equation (30), (31) and (32) with the values from table l one finds the refraction
angles in (sexagesimal) seconds of arc per kilometre in table l l.
*) It may be useful to measure q on a visual wavelength and to determine A q photo-electrically OIJ two

other wavelengths far apart.

PUBLICATIONS ON

GEODESY, NEW SERIES, VOL. 3,

NO.

Table 11. Refraction angles and humidity corrections

1
refraction angle

(v1- H)/S

1 humidity correction C.

typical

high

( + 12 "/km
1 +0.02"/km 1 +0.8 ''/km
-4

"/km

1
I-

low
50"/km

- 1"/km

If it becomes possible to measure the dispersion angle A? with sufficient accuracy, the
refraction angle ( v , - H ) can be calculated. The accuracy will then be limited by the refraction from humidity C, or by inaccuracies of the dispersion formulae if the wavelengths
are chosen too far from the visual region.

Chapter 8
CONCLUDING REMARKS

If the instrumental errors are kept sufficiently small and if two wavelengths are used, a gain
in the precision of E.D.M. amounting to one order of magnitude should be possible. If
additionally a radio wavelength is used to correct for the humidity, essential gain will only
be reached in a very humid environment.
If the formulae for the refraction index of visual- and of infrared light should be better
known, a somewhat better accuracy is probably possible with two optical wavelengths.
With three wavelengths at least a gain of one order of magnitude should be possible.
The dispersion method may also be used to measure a refraction angle, but in this case
it is extremely difficult to obtain a sufficient instrumental accuracy, because the angle
measurement is very sensible to local changes in the refraction index, and because the
dispersion of the air for the phase refraction index is only one third of the "group dispersion". Physically the accuracy is limited by the gradients of the humidity to 0.02"/km in
normal conditions. This limit however may be decreased if the humidity gradient is measured
or if the angles are measured from both ends.

Addendum
When this paper was ready in draft, I received the very interesting ESSA-report of Mr.
THAYER
[16], covering about the same subject. In this report Mr. THAYER
considers distances
between ground stations too, but more in detail, distances to satellites. I did not consider
the latter cases although my approach may well be applied to distances to satellites.
The general conclusions of Mr. THAYER
about terrestrial distances are in good agreement
with my conclusions, in detail however there are some differences. In general my assumptions about meteorological circumstances are more "pessimistic" than Mr. THAYER'S.
The error caused by the difference in the temperature-function of water vapour and dry
air (my 8,) is in the report of Mr. THAYER
a factor 2.5 smaller than the value given in my
paper. The cause must lay in the difference between his humidity formulae (32) and mine.
I could not trace this difference in detail.
finds in his article [4] a smaller dependence on the humidity than I do. He
Mr. TENGSTROM
however does not account for variations of the humidity gradient, which variations are
in my opinion more important than the effect mentioned by Mr. TENGSTROM.

REFERENCES

[ l ] H. MORITZ- Zur Reduktion elektronisch gemessener Strecken und beobachteter Winkel wegen
Refraktion. Z. Vermess.-Wes. 86 (1961), pp. 246-252.
[2] M. NABAUER
- Terrestrische Strahlenbrechung und Farbzerstreuung. Bay. Akad. Wiss. Miinchen,
19291111.
[3] J. J. LEVALLOIS,
G. DE MASSON
D'AUTUME
- Etude sur la refraction geodesique et le nivellement barometrique. IGN, Paris, 1953.
141 E. TENGSTROM
- Elimination of refraction at vertical angle measurements using lasers of different
wavelengths. Proc. Int. Symp. Vienna, 1967. ijst. Zt. Vemess., Sonderheft 25 (1967), pp. 292-303.
[51 J. C. OWENS- The use of atmospheric dispersion in optical distance measurement. Bull. Geod. 89
(1968), pp. 277-292.
- Refractometer for determining the mean
[6] M. T. PRILIPIN,A. N. GOLUBEV,L. M. KONOSHENKO
refractive index of air in pulsed-light range finder measurement. Geod. Aerophot. (USSR), 1968,No. 2,
pp. 62-67.
[7] F. E. TERMAN
- Radio Engineers' Handbook. MacGraw Hill, New York, 1943.
[8] 13th Gen. Ass. U.G.G.I., Berkeley, 1963, Resolution 1. Bull. GBod. 70 (1963), p. 390.
[9] M. BORN,E. WOLF- Principles of Optics. Perg. Press. Oxford, 3rd ed., 1965.
[l01 J. C. OWENS- Optical refractive index of air: dependence of pressure, temperature and composition.
Appl. Opt. 6, 1 (1967), pp. 51-59.
[l l ] B. EDLBN- The refractive index of air. Metrologia 2, 2 (1965), pp. 71-80.
[l21 The refraction index of air for radio waves and microwaves. Nat. Phys. Lab. Teddington (G.B.), 1960.
[l31 J. SAASTAMOINEN
- Curvature correction in electronic distance measurement. Bull. GBod. 73 (1964),
pp. 265-269.
[l41 W. BISCHOF- Periodical variations of atmospheric CO,-content in Scandinavia. Tellus 12 (1960),
NO. 1, pp. 216-226.
[l51 LANDOLT/BORNSTEIN
- Zahlenwerte und Funktionen Band 111, 6th ed., Springer, Berlin 1952, section 32821.
[l61 G. D. THAYER
- Atmospheric effects on multiple frequency range measurements. E.S.S.A. Technical
Report IER 56 - ITSA 53. Boulder, (Col.) U.S.A., (1967).
[l71 M. V. R A ~ N S K-I The
Y problem of determination of the index of refraction of the air during distance
measurement with electro-optical and pulsed-wave range finders. Geod. Aerophot. (U.S.S.R.), 1962,
No. 2.
[l81 New values for the physical constants. Techn. News Bull. ,Nat. Bur. Stand. 47, No. 10 (19631, pp.
175-177.
[l91 W. HOPCKE
- On the curvature of electromagnetic waves and its effect on measurement of distance.
Surv. Rev. 18, No. 141 (1966), pp. 298-312.
[20] R. BREIN- Die Bestimmung der atmospharischen Refraktion aus der Dispersion des Lichtes, Deutsche
Geod. Komm., Reihe B, Heft Nr. 165 (1968).

Table 1. Quantities about the refraction index

Functions of the wavelength


(dispersion) quantity

curvature of the earth R


(meteorological)
quantity
(lengths in
metres)

1
l

6.38.10V m

I
I

Functions of meteorological conditions

light
typical

high

low

microwaves
typical

high

low

Table 2. Explanation of the terms of equations (respectively table 3, 4, 5)

I terms of equation (19a) (table 3)


1st
2nd
3rd
4th
5th

first order correction for the refraction index of dry air


first order humidity correction
dry air curvature correction
mixed curvature correction
humidity curvature correction

1 terms of equation (2la) (table 4)


1st
2nd
3rd
4th

first order humidity correction


dry air curvature correction
mixed curvature correction
humidity curvature correction

1 terms of equation (27a) (table 5)


dry air curvature correction
mixed curvature correction
humidity curvature correction

. . . .

-22I

(D

(D

i i i i

(D

2. 2. 2. 2.
N O O N
N

2. 2. 2. 2.

- 0 0 -

9 N.

91

I I I I

I I I I

. . .

. . . .

++++

d 0 0 d

++++

m o o m

'

------ b-. b-. b--. b-. b-. b--.


-

m m m w ( D ( D

bbbbbb
. . . mI I .
m

=?c?
9*
9 1 9
o o m o o m
0-00-0
+-t++++ 1 1 1 1 1 1

Table 4. Meteorological relative corrections on two optical wavelengths

No.
1st

1 S in km l

term of equation (21a)

$10

- ( F l - DdLF)eL

2nd

s2-

{
3rd

12
dG d f

-1

dg dg
--

ae

ae

s 2

az az

T " , - D A ~ P ae
12R

4th

24

1
3
10
30
100
300

-0.06~10-s
-0.6.10-0
-6
.10-O
-0.06.10-"
-0.6 .lO-e
-6
.10-"

1
3
10
30
100
300

-0.3 .10-l2
-3
.10-l2
-0.03.10-8
-0.3 .10-O
-3
.lO-O
-0.03.10-"

24

24

high

typical

az

s2

.10-"

Table 5. Meteorological relative corrections at two optical- and one micro wavelength
NO.

term of equation (27a)

12

-K,A,{

dG dl-dg dg ae

ae
az az

- KLAL

typical

1
3
10
30
100
300

+0.3 .10-l2
+3
.10-l2
+0.03.10-0
t0.3 .10-8
$3
.10-0
t0.03.10-B

s2-

p
-

2nd

I S in km 1

dG dl-

-I

--

dg dg
12

ae ae
az az

S,

high

low

temperature
total pressure
relative variation of dry air density

temperature for the 3-l-method


CO, partial pressure

humidity gradient

dry air refractivity gradient

humidity

maximum vapour pressure at air


temperature
vapour pressure

quantity or its variation

i;

at,, a t ,
apco2

aelaz

s(ae/az)

Psmax

6t
6p
6e/e

0.5 "C
0.04 torr

6.5 torr

0.5 "C
0.7 torr
1.7.10-~

favourable

0.04 torr

13 torr

1 "C
0.7 torr
3.3.10-3

typical

0.2

-1.3. 10~-3
m-'

72 torr
.
P
0.1
0.1

2 "C
1.5 torr
6.7. 10-3

unfavourable

6tQ = a t ,
1141

= 6t

own estimate

(see above)

from table 1 : 50% of the typical value, 100% of


this value, or the most unfavourable value

calculated from 6p3 with appendix I

for 5 "C, 15 "Cand 45 "C


~151 [l71~
~

[l71 and own estimates


[l71 and own estimates
calculated from 6t and 6p above with appendix I

source

Table 6. Assumed values for the inaccuracies of meteorological measurements and estimations (to be interpreted as standard variations)

0.3

.10-6

[ l l , p. 76, under (l),


p. 79,
table 6, X 3 for group-effect]

[l 1, p. 73, first column]

literature

[l81

1
1

I
.10-6

0.3

.10-6

0.005. 10-g

0.05 .10-'

0.08

0.5

value

1
1

[l81

[10, section I, p. 511, [l l ]

[lO, section I, p. 511, [l11

[10, section V, p. 581


(0.25% of d ~ i i )

110, table v ]

literature

unfavourable conditions

*) The difference between the formulae (12) and (15) of [l11 appear to have no significant influence on the values of this table.

I light velocity in vacuum

1 ESSENand FROOME

[10, section I, p. 511, [l11

value

favourable- or typical conditions

[10, table 1111

6c/c

[10, section I, p. 511, 1111

deviation

OWENS

EDLEN*) [l 1, equation (l), (15)


and (22)]

formulae

Table 7. Inaccuracies of the refraction formulae for air and of the light velocity in vacuum

1
1

-4
oo

328

ggg -

l o

I2

8
to

pp'-'

l.;?

0 0 0 0

0 0

0 0 0

0 0

--C-,

3 3 3 3

0 0 0 0

0 0 0 0

$8

vv

45e

C)

9 ':
0ol.y

l 0

2p
2

0 0 0 0

-?a;,

g
2P
2

- m

- m

88sq
0 0 0 0
v

88em
0 0 0 0
v

82

-+m

888s

oooo

8
-?
0 o *

bn
C
.3
u

--

cd

E
.bn
.
5

$ 0

od
Q)

vv

0 0 0 -

0000

0000

mngx

mngg

82

00

gg
=?

88

PP

ln

vv v

88

00

g "?

;?

m m
1

NQ

ln

" 9

0 0
nn
4

8888
0000
V v

oo

00

8
;?
9 "?
oo

rzmm

2~

- M

Wc4
.

h
.d

a
c
h h

lnw

33

---l.

00

ln

00.N

!g;?
l

l
.

-8 8- 8- 8

oooo

gg8T

C
V)

00

33

10

C 5

S .-

2z

-8 8- 8- 8

V)

&

82

m Q

pp

888s
v v

0000

A 0 l

- - W

6000

- 0

0 0

S
a,
0600

5 g849

252
0 0 0

a"?

N X 2X*cO

4d

vvv

338

2%

8888

0 0 0 0

-* g k s
.3

vvv

g?TP

V V

Ng

gggg 58

l
.

0 0 0 0

8888

a
a
C

82

8 2 N m g0 80 -090

m;?

3g

.-3 3
S

G
!

,F?

&

CH
e
WO

Q,Q,

:S
-G&

wog

-.-8
C)

"E, '

83

Appendix I
S O M E VALUES USED I N T H I S P A P E R

1 General

For the refraction index of atmospheric air two sets of formulae are used. For visual light
the results of EDLBN[l l , equations (12), (1) and (22)] are written in the form (1.a) and (1.c).
For radio waves the formula of ESSENand FROOME
[12, equation (l)] is used in the form
(1.b) and (1.d):

where:

n,
g
p
p,
t

and nM are the (phase) refraction indices for visual light and for radio waves,
is the wave number ( = llwavelength) in (pm)- l
is the total pressure in torr,
is the partial pressure of the water vapour in torr, and
is the temperature in centigrades.

The arbitrary factors "1.2168" and "72" in ([.c), and the numerical values of GM and T M
in (1.d) are so chosen that the quantities q and e will hardly become larger than unity.

40

PUBLICATIONS ON

GEODESY,
NEW SERIES,
VOL. 3,

NO.

2 The factors independent of the meteorological conditions (G, T,etc.)


G, and r, have been calculated for g , = 2.75 (pm)- ' and for g, = 1.6 (pm)- ' (near the
limits of validity of the E D L ~ expressions).
N
GM and T, are directly given by (1.d). See
table 1 of this paper.
c,, F, and the other functions of g are also calculated from (1.c). The analogue functions
for microwaves are easily found because there is no dispersion in this case.
These values are also assembled in table 1 of this paper.
3 The factors depending only on local meteorological conditions (Q,, Q,, e,, e,)

Assuming the meteorological conditions *) of table 1.1 typical, very high and very low
values of the quantities Q and e have been calculated for light and for microwaves. The
results are assembled in table 1 of this paper.
Table I. 1. Assumptions for meteorological conditions

typical

high

low

quantity :
Q,,Q,
e ~e~,

+14"~760torr- - 3 0 "
+ 14 "C 760 torr 60% +45 "C

typical

high or low

4 The factors depending on the vertical variations of the local meteorological conditions
(functions of a ~ / a Z ae,/az
,
and ae,/az)

The vertical derivates of Q and e are found by differentiating the third and fourth expressions
of (1.c) and (1.d). With good approximation the following expressions are found:

*) The assumptions are made for terrestrial measurements. For much greater heights (aeroplanes, satellites)

the values of

and e tend to become zero, provided ionospheric effects are unimportant.

41

THE THEORY OF DISPERSION, ETC.

In order to obtain numerical values for de/dZ, aeL/aZand ae,/aZ use is made of the results
[l91 who gives a great number of values for the refraction coefficients kL and k,
of HOPCKE
which have been calculated from meteorological measurements of at/aZ and ap,/aZ.
HOPCKE
used the formulae (1.f):

assuming:

E here expressed in kL
With (1.f) and (1.g) the basic meteorological values of H ~ P C Kare
and k,:
at
- - - 0.155kL+0.001kM-0.034 . . . . . . . . . . . . . . . . . . (1 h)

az

and :
JP,

--

az

0.036kL-0.026kM-0.002

. . . . . . . . . . . . . . . . . .

(1.i)

Numerical values of the terms of the equations (19a), (19b), (21a), (21b), (27a) and (27b)
are found by using the equations (I.e), (I.g), (1.h) and (1.i) with the meteorological conditions of table 1.1. So one finds:

for all values of the dry air terms and


for the typical values of the humid- and mixed terms

for high- or low values of the humid- and mixed terms

42

PUBLICATIONS ON

GEODESY,
NEW SERIES,
VOL. 3, NO. 4

Values for k, and kM were estimated from the graphs of HOPCKE[15]. For the typical
values was chosen :
k,

+0.25

and

kM = +0.30

. . . . . . . . . . . . . . . (1.k)

For high and low values, combinations of k , and k, were chosen for each individual term.
A few examples are indicated in table 1.2.
Table 1.2

+1
+2.5

-1

+1

some combinations of k, and kM from the graphs


of HOPCKE[19], giving high- or low values of
different terms of (19a), (19b), (21a), (21b), (27a)
and (27b)

5 The relations between the meteorological factors for radio waves and for light waves
(6, and 6,)

In chapter 5 of this paper a comparison is made between measurements on optical- and


radio wavelengths. In this case the ratio's 8, = eM/eLand 8, = eM/eLare wanted in some
approximation. These values are easily found from (1.c) and (1.d):

Appendix I1
T H E L O R E N T Z - L O R E N Z EQUATION F O R
T H E REFRACTION I N D E X

H. A. LORENTZ
and L. LORENZ
have derived a formula for the refraction index of a medium
[9, chapter 111. For one component this equation may be written as:

where G is only dependent on the (vacuum) wavelength and Q on the density, i.e. on temperature and pressure. If n - l << 1 this expression does not differ much from an expression
such as (9) for dry air:
n-l

=GQ

. . . . . . . . . . . . . . . . . . . . . . . . . .

(1I.b)

Neglecting higher powers of Ge one finds for dry air from (1I.a) and (1I.b) and with (16):

In this paper the formula for the refraction index of dry air was supposed to be of the
form (II.b), which form is often used, for example by EDLBN[l l]. If the refraction index is
calculated with the probably more accurate Lorentz-Lorenz expression (II.a), (OWENS
[IO]), one should apply some correction to the equations to calculate the distance S. Since
the influence of the humidity is normally only small and difficult to introduce into the
calculations, only the dry air correction is given here.
The corrections are found by substituting Ge and Ge with (1I.c) into (19a), (21a) and
(27a). Also in the quantities D, K, and KM this substitution should be made.
So one finds forms analogue to (19a), (21a) and (27a), but with CS and &j instead of
Ge and Ge (also in ALG, D, KL and KM),and with an additional correction term 6,, to the
right-hand member of (19a), (21a) and (27a):

Appendix 111
THE ERROR INTRODUCED I F AN
ELECTROMAGNETIC DISTANCE MEASUREMENT IS CALCULATED
WITH T H E G R O U P REFRACTION I N D E X

The usual way of deriving a distance S from the measured optical path a, is principally
different from the theory in this paper: normally one substitutes the group refraction
index fi (defined by (7)) in the monochromatic solution of the Maxwell equations instead
of substituting the phase refraction index n, and introducing the group effect later on.
This usual method however is not more than an approximation because the Fermat principle
(8) does not hold generally for the group refraction index. In order to demonstrate the
(very small) error, the "group path" a, will be calculated. This quantity is defined by the
substitution of f i instead of n in the monochromatic solution for a.
In the used approximation one finds this a, by substituting fi, afi/aY and aii/aZ in (12a)
for the corresponding n-values. From (7), (9) and (16) one finds:
E = l+Ge+Pe

aii
ay

-=

Ga@+ r- -ae
ay

ay

afi

--

az

a@ + l-- ae . . (1II.a)
G--az
az

Substitution in (12a) and (13) gives :

Comparing this form with (14) and (15) one finds for the error of the usual method of
calculation :

With the values of table 1 this error is found to be smaller than 0.03 Q. Since the Q-terms
can only very roughly be measured or estimated the error (ag-a,) can be neglected in all
practical cases.
In the formulae (19b), (21b) and (27b) of this paper (ag-a,) is neglected. The forms
become so much simpler, but the influence of the humidity does not appear explicitly.

Appendix IV
THE INFLUENCE O F ERRORS
I N E S T I M A T I N G T H E F I C T I T I O U S T E M P E R A T U R E t,
(See chapter 5 under equation (23b) and section 6.2, "row 5 and 6")

To find the influence oft, the equations (25) and (26) are partially differentiated, neglecting
the higher order terms. After some calculations one finds:

aKL
- -

KMr"M.(eeGM- G')
ALG.(~,TM
-Fl) - A~T.(o,G, - G,)

aKM -

- K M f M.(ALG)

ae,

ae,

A~G.(B,P,-Pl) - ~,r".(e,G, - G,)

Using the model (17) of chapter 3 one can substitute (18) in the second and third equation
of (24). So one finds with neglection of the higher order terms:

From the above expressions one easily finds:

as

-=

sat,

do,
KMr",eL-dt,

In the same way:

sat,

d 0, G
~ -d te

Using the approximation (23b) for 9, and for g,, the influence of t, should be neglected,
and for t, one finds with the values of table 1 :

typical conditions

high humidity

With the inaccuracies for t, stated in table 6 the values of row 5 in table 8 are found.

46

PUBLICATIONS ON

GEODESY,
NEW SERIES,
VOL. 3, NO. 4

The above mentioned results are derived from formula (23b) of this paper which is based
N l]. In these formulae the humidity correction is independent
on the formulae of E D L ~ [l
of the temperature. However elaborating the Owens' formulae [l01 numerically, one finds
there a humidity correction essentially inversely proportional to the absolute temperature.
For a wavelength of 0.625 pm one finds for e, from OWENS:

So one finds instead of the second equation (23b):

The influence of t, for the Owens' formulae appears to be about half the values given in
row 5 of table 8.

Appendix V
THE INFLUENCE O F ERRORS I N THE GROUP
R E F R A C T I O N I N D E X (Sfi,, SE,) A N D I N T H E D I S P E R S I O N (6ALfi)

1 For one wavelength

For this case (19b) will be applied neglecting the higher order terms:
S

= a,+S(l

-fi)

The differential form gives the error 6S caused by an error 6fi in the group refraction index:
SS

or:

6s
S

(1 -fi)6S-S6fi
62
fi

2 For two optical wavelengths

Neglecting the higher order terms in (21b) one finds


S

= a,, - DALa,-(fi,

- 1)S+ DALES

The differential form gives the error 6 S caused by the errors S f i , and 6ALfi:
6 S = - A,a,GD - ( f i ,- 1)6S+ DALfi6S-SGfi,

+DSALfi+ALfiS6D,

where 6 D is the variation in D caused by variations in f i , , and in ALE, from which follows
that the coefficient of 6 D becomes approximately zero.
It may easily be seen that:
ALa, x SALE

So one finds:

(following from (14), (II1.a) and (18)

48

PUBLICATIONS ON GEODESY, NEW

SERIES,
VOL. 3, NO. 4

or with good approximation:

3 For two optical- and one radio wavelength


Neglecting the higher order terms in (25) and multiplying its two members with:

one finds:

In the differential form of this equation for constant a,,, ALa, and AMa, one substitutes:
a,,

S.(I +G,eL+r",eL)

ALam= s . ( A ~ G - ~ ~ + A ~ ~ " - ~ ~ )
AMa,

s'(GMeM- GleL+r"MeL-F1eL)

(following from (14) and (18), neglecting the higher order terms)
So one get after some elaborations: